Intel Emerald Rapids CPUs are up to 2.4x Faster than Sapphire Rapids in AI/HPC Workloads

Intel’s 5th Gen Xeon Emerald Rapids processors are set to launch on the 14th of December, alongside Sierra Forest and Meteor Lake. A soft refresh of Sapphire Rapids, they’ll offer slightly higher core counts, a lot more cache, and faster core/memory clocks than its predecessor. Structurally, it’s a regression from Sapphire, consisting of two large chiplets, down from four on 4th Gen.

Emerald Rapids will offer up to 64 cores and 480 MB of L3 cache, higher than the corresponding AMD Epyc parts. The L2 cache adds up to 128MB for a total of 608 MB. Each chiplet comprises 35 cores (3 disabled), dual memory controllers supporting up to DDR5-5600MT/s DIMMs, and three PCIe controllers.

Intel claims a generational uplift of 20-30% with Emerald Rapids in data center workloads, ranging from web, HPC, and media. Specialized accelerators boost AI (AMX), network (QAT), streaming (DSA), and database (IAA) throughputs by up to 240% over preceding Sapphire Rapids offerings.

Intel’s 5th Gen Xeon family will consist of four stacks divided into the P and E-core segments and further into the SP (Scalable Performance) and AP (Advanced Performance) tiers. The P-core lineup will use the Raptor Cove core architecture, while the E-core variants leverage Sierra Glenn, a Crestmont variant optimized for cloud workloads. The SP and AP stacks will vary in core counts, cache, and IO capabilities, the former addressing the lower-end and the latter addressing the higher-end segments.

Via: InstLatX64.

Areej Syed

Processors, PC gaming, and the past. I have written about computer hardware for over seven years with over 5000 published articles. I started during engineering college and haven't stopped since. On the side, I play RPGs like Baldur's Gate, Dragon Age, Mass Effect, Divinity, and Fallout. Contact: areejs12@hardwaretimes.com.
Back to top button